Mac OS X下用sed将BLAST XML中OS=后内容全部放入[]的方法

The issue with your original sed command is that the character set you used ([A-Za-z\.\-\,\_ ]) doesn't cover special characters like colons, slashes, or parentheses. Plus, relying on a fixed character set isn't flexible for the varied organism names in BLAST outputs.

Here's a robust, two-step sed approach that works with macOS's BSD sed:

sed -E 's/(<Hit_def>.*)OS=/\1[/' input.xml | sed -E 's/( [A-Z]{2}=)/]\1/' > output.xml

How it works: #

1. First sed command: Finds the <Hit_def> tag content and replaces OS= with [, effectively starting the bracket around the organism name.

   • (<Hit_def>.*) captures everything from the start of <Hit_def> up to OS=
   • \1[ replaces that captured text plus OS= with the captured text plus [

2. Second sed command: Locates the first occurrence of a field marker (like OX=, GN=, etc. — all follow the pattern [A-Z]{2}=) and prepends a ] to close the bracket around the organism name.

   • ( [A-Z]{2}=) captures the first field marker after the organism name
   • ]\1 replaces that captured marker with ] plus the marker itself

Example test: #

For your problematic input line:

<Hit_num>1</Hit_num> <Hit_id>TR:V4U0L5_9ROSI</Hit_id> <Hit_def>gnl|V4U0L5| Uncharacterized protein OS=Yersinia enterocolitica serotype O:8 / biotype 1B (strain NCTC 13174 / 8081) OX=393305 GN=CICLE_v10008136mg PE=4 SV=1</Hit_def> <Hit_accession>TR:V4U0L5_9ROSI</Hit_accession> <Hit_len>388</Hit_len>

This command will output:

<Hit_num>1</Hit_num> <Hit_id>TR:V4U0L5_9ROSI</Hit_id> <Hit_def>gnl|V4U0L5| Uncharacterized protein [Yersinia enterocolitica serotype O:8 / biotype 1B (strain NCTC 13174 / 8081)] OX=393305 GN=CICLE_v10008136mg PE=4 SV=1</Hit_def> <Hit_accession>TR:V4U0L5_9ROSI</Hit_accession> <Hit_len>388</Hit_len>

This method handles all special characters in organism names because it doesn't rely on matching specific characters — it just targets the start (OS=) and end (first field marker) of the organism name segment.

内容的提问来源于stack exchange，提问作者adee1212